Adata unveils "the world's fastest M.2 SSD"

midian182

Posts: 6,075   +50
Staff member
What just happened? We knew that PCIe 4.0 would lead to consumer SSDs with blistering speeds, and they're starting to arrive, including what is claimed to be the fastest M.2 SSD in the world—at least when it comes to read speeds.

The XPG Gammix S70, which comes from Adata, utilizes the PCIe 4.0 interface to reach sequential read speeds of 7,400 MB/s and sequential writes at 6,400 MB/s. That’s faster than the current speed leader, the Sabrent Rocket 4, which is rated at 7,000 MB/s—the same as the upcoming Samsung 980 Pro PCIe 4.0 M.2.

The XPG Gammix S70 isn’t number one when it comes to write speeds, though, as Sabrent’s SSD beats it by just 450 MB/s. Samsung’s product is slower, at 5,000 MB/s writes.

Adata’s SSD boasts XPG’s CoolArmor aluminum heatsink that uses a terraced structure design to increase its surface area for improved heat dissipation. Combined with the hollow chambers underneath for better airflow, the heatsink can reduce load thermals by up to 30 percent.

The drive offers plenty of other features: LDPC error correcting, AES 256-bit encryption, RAID engine support, Dynamic SLC Caching, and DRAM Cache Buffer. You even get a 5-year warranty.

Adata says there will be 1TB and 2TB versions of the XPG Gammix S70, though we still don’t know pricing or availability. Remember that to get the top speeds, you’ll need a motherboard that supports PCIe 4.0, which right now is limited to AMD’s X570 and B550. The SSD is backward compatible with PCIe 3.0, but you’re not going to get the same level of performance, of course.

Permalink to story.

 

trparky

Posts: 814   +781
For that you need Intel Optane SSD...
I'm referring to the fact that many of these SSDs advertise huge sequential benchmark speeds, but they hide their 4K random numbers. Why, do you ask? Because generally speaking when it comes to 4K random benchmarks which is the number you really should be paying attention to (since that's the number that most closely mirrors that of what you'd expect from it in the real world) you'll find that it doesn't matter what SSD you get.

Just about any SSD will give you the performance that you want, everything else is simply marketing BS to make you think that it deserves a higher price.
 
  • Like
Reactions: killmess

Gomos

Posts: 32   +59
I'm referring to the fact that many of these SSDs advertise huge sequential benchmark speeds, but they hide their 4K random numbers. Why, do you ask? Because generally speaking when it comes to 4K random benchmarks which is the number you really should be paying attention to (since that's the number that most closely mirrors that of what you'd expect from it in the real world) you'll find that it doesn't matter what SSD you get.

Just about any SSD will give you the performance that you want, everything else is simply marketing BS to make you think that it deserves a higher price.
"The real world" is different for every user. Workloads that require huge sequential speeds do exist.
 

Uncle Al

Posts: 7,485   +5,992
I have read some rather nasty comments about Adata on Amazon (take that for what it's' worth) with a number of complaints about Adata products failing on start up to about two months of use. No comments on the companies warranty or return policy so I'm going to have to give it a good long wait before I invest, despite it's products tend to be just over half of the big boys!
 

ferrellsl

Posts: 44   +41
"The real world" is different for every user. Workloads that require huge sequential speeds do exist.
You are correct. I work with very large LIDAR point clouds, some with billions of points and file sizes in excess of 6GB. This is a very welcome advance in read/write speeds for people in my field.
 

kiwigraeme

Posts: 122   +88
As others have already alluded to - I do video encoding - is nice that M2 drives can analyse a 30Gb video in very few seconds -much much faster than a SSD drive
 

Dyper

Posts: 22   +8
I love NVMe M.2’s for loading my Samples. No bench for that workload but it is based on time, which is only an extra second or two, but that’s enough for me.

My guess is this is using the Phison E18.

Hope the reviewers get a freebie for their praise, then do a thorough review.

 

flipp3r

Posts: 26   +5
I have read some rather nasty comments about Adata on Amazon (take that for what it's' worth) with a number of complaints about Adata products failing on start up to about two months of use. No comments on the companies warranty or return policy so I'm going to have to give it a good long wait before I invest, despite it's products tend to be just over half of the big boys!
From personal experience their SX6000 range is really bad with early firmware. The drives randomly disappear. To update the firmware you need to have them as a secondary drive.
The SX8200 has been good. Both have different Realtek Chipsets...
 

Evernessince

Posts: 5,418   +6,010
From personal experience their SX6000 range is really bad with early firmware. The drives randomly disappear. To update the firmware you need to have them as a secondary drive.
The SX8200 has been good. Both have different Realtek Chipsets...
The SX8200 is the best gaming SSD on the market, boasting latencies only matched by an HP drive with the same controller.

Only Intel optane has lower latencies but that drive comes with it's own set of drawbacks, aside from insane price.
 

jpuroila

Posts: 234   +127
I have read some rather nasty comments about Adata on Amazon (take that for what it's' worth) with a number of complaints about Adata products failing on start up to about two months of use. No comments on the companies warranty or return policy so I'm going to have to give it a good long wait before I invest, despite it's products tend to be just over half of the big boys!
I've had SU800 for over three years now and have no complaints about it, but you shouldn't mistake anecdotes for data.
 

Squid Surprise

Posts: 3,453   +2,348
If by "from" you mean "for", then isn't that what he said?
Yes... typo on my part... fixed... but no, he didn't say that... or at least, that's not what his post implied. He was replying to someone who said the drive wasn't good by saying that in his experience, the drive is great...

But to actually state that the drive is great, you need actual data - your own personal experience is statistically irrelevant.
 

Endymio

Posts: 623   +514
Yes... typo on my part... fixed... but no, he didn't say that...to actually state that the drive is great, you need actual data - your own personal experience is statistically irrelevant.
What you think he meant by, "you shouldn't mistake anecdotes for data"?
 

Endymio

Posts: 623   +514
His coordinating conjunction, "but", makes the most sense when applied to his own sentence, not someone else's.
 

Gomos

Posts: 32   +59
I understand that however for 95% of users out there, any SSD is going to be better than a spinning rust drive.
Sure, maybe even 99%, but the remaining 1% are likely extremely important users for the advancement of science and technology.